Yes! – “The Secure Health Informatics Research Environment (SHIRE) is a HIPAA-compliant, secure cloud computing platform hosted by UNC Health that provides flexible compute and analytical tools for data-driven research projects. Data are retained in project-specific workspaces with controlled access. Firewall rules are set to prevent any data exfiltration unless explicitly allowed.”
If you need more detail, you may want to reference the SHIRE Fact Sheet.
User Account Questions
Yes. You can add additional IRB-approved users by submittingthis form at any time.
You can remove a user by by submittingthis form. Prior to submitting this form, you should (1.) ensure the user’s VMs do not have any key data or code and (2.) ensure all of the user’s VMs have been deactivated.
Yes—these trainings are required as part of having UNC Health login credentials.
Technical Questions
We are happy to review any user requests for software to be added to the SHIRE, using our software request form. However, licensed software for which we do not have an institutional license may be difficult for us to support.
For example, Microsoft Office products and SAS are licensed software for which UNC has an institutional license we are able to support them in the SHIRE.
On the other hand, STATA and SPSS are licensed software for which UNC does NOT have an institutional license. We do NOT currently have a way to support those tools.
Likely yes, but it depends on a few factors. If you received de-identified data from the CDWH and trained your model on those data, you can request extraction of the model file as a Type 1 export—see our Export request documentation for more details. If you trained your model on HIPAA-limited or identified data, you may still be able to export the trained model if you are able to prove you have taken steps to prevent identifiers from leaking into the model (by, for example, de-identifying the dataset). Requests to export models for HIPAA-limited or identified data requests need approval by the CDWH Operations Committee. To request approval, please contact the TraCS analyst or PM you worked with.
The SHIRE automatically shuts down all Virtual Machines (VMs) at 7:00 PM Eastern Standard Time each night unless you specifically “snooze” or cancel the shutdown via the links in that email. The auto-shutdown is in place to prevent users from accidentally leaving VMs on perpetually and running up unintended charges.
Members of the SHIRE team have modeled several different parameters surrounding LLM utilization to assist users in budgeting and cost estimation. This modeling may allow researchers to better estimate their anticipated costs based on a variety of factors. However, users should know that it is impossible to fully predict cost of LLM usage with more than moderate precision.
Billing Questions
Yes. Please reference the Quick Billing Model Guide to learn more about the line items that may appear on your weekly cost report.